THERMAL PROPERTIES OF LIVING ROOF GREENING MATERIAL BY INVERSE MODELING

Abstract

Sunagoke moss mat is a composite material with good potential for active roof greening in urban heat island mitigation. However direct measurement of its thermal properties is difficult since it is made up of organic (live Sunagoke moss and cotton wool) and inorganic (PVC netting) constituents. Thermal conductivity, thermal resistance, and universal heat transfer coefficient of a Sunagoke moss mat were determined using inverse modeling and neural network optimization. The research hypothesis was that these properties can be modeled as weighted averages of corresponding properties of the composite’s constituents. Temperature regimes on either surface of a Sunagoke moss mat (dimension 110 × 100 × 18.1 mm) in an insulated thermal transmission system were measured at different volumetric water contents. Temperature regimes at 0 and 100 volumetric water content were used as input for a computer algorithm to compute thermal conductivities at various temperatures. A weighted average thermal conductivity model was developed and a Multi-Layer Perceptron (MLP) neural network with one hidden layer was trained to optimize its coefficients. The thermal resistance of the material was 14.15 and 5.08 KW-1, respectively, at volumetric water contents of 0 and 100%. The universal heat transfer coefficient was 6.42 and 17.9 Wm-2K-1, respectively, for the same conditions. Both of these properties varied with volumetric water content.